Explaining models predicting players market value with permutation variable importance.

I am going to explain models predicting players market value using permutation variable importance. I use fifa-23 dataset. Models that I use are Random Forest Regressor, CatBoost Regressor and MLP Regressor.

First model that I use is RandomForestRegressor. I have already used and shortly described in homework 1.

MLP - multi-layer perceptron is a neural network that consists of at least 3 layers. One layer usually consist of a linear and an activation.

Third model is Catboost Regressor. The main idea of boosting is to sequentially combine many weak models and thus through greedy search create a strong competitive predictive model.

Here I used Random Forest again but with normalized data. As expected there is no much difference between previous usage.

  1. Most important variables are 'Reactions', 'Stamina', 'BallControl' and 'ShortPassing'.
  2. MLP's explanation differs from other models that are tree-based.